NVIDIA expands CUDA-Q with orchestration layer for fault-tolerant quantum systems

NVIDIA expanded its open-source CUDA-Q platform with CUDA-Q Logical, an orchestration layer for programmable and verifiable development of applications for fault-tolerant quantum computers.

Researchers can combine algorithms, error-correction codes, hardware architectures and other components, then compare configurations by resource needs and performance. Such systems are intended for applications including drug discovery, financial modeling and materials development.

Fermilab cuts development to three weeks

Fermilab used CUDA-Q Logical to evaluate error-correction approaches, physical qubits, runtimes and other resource requirements. Fault-tolerant architecture development fell to three weeks from about five months, a sevenfold acceleration.

Iceberg Quantum modeled an architecture for Diraq’s qubits and estimated that 1,000 logical qubits could be created with 150,000 physical qubits—roughly ten times fewer than Diraq’s previous estimates.

QUOPS measures hardware readiness

Sandia National Laboratories added QUOPS to CUDA-Q. The open, cross-platform benchmark measures progress toward practical fault-tolerant quantum computing and is hardware-agnostic. It complements conventional measures such as physical-qubit counts, fidelity and coherence time.

Sandia reported initial QUOPS results for quantum processing units from Google, IBM and Quantinuum in a preprint released ahead of IEEE Quantum Week. A QUOPS reference implementation is available in CUDA-Q.

CUDA-Q Logical is already being used or evaluated by Fermi National Accelerator Laboratory, Infleqtion, IQM Quantum Computers, QCDesign Quantum Motion and Sandia National Laboratories. CUDA-Q Logical and the QUOPS repository are available through GitHub.